France’s Tesla FSD Road Test Is an Independent Check Before a Possible EU Vote

France has started testing Tesla’s Full Self-Driving (Supervised) system with two cars on French roads. The move follows France’s earlier opposition to authorizing the system in its then-current form, citing concerns about its ability to exceed speed limits and the adequacy of driver-attention safeguards. The practical purpose is to give French authorities their own evidence before a possible European Union Technical Committee on Motor Vehicles vote on October 6. 1718

What France’s trial is designed to establish

The French testing is best understood as a regulatory cross-check. It can show how FSD (Supervised) behaves on French roads and in French traffic conditions rather than asking France to rely solely on Tesla’s evidence or the Dutch road authority RDW’s assessment.

That matters because the Dutch RDW granted a provisional approval after 18 months of testing on tracks and public roads, then notified the European Commission of its intention to seek wider use. Belgium, Denmark, Estonia and Lithuania subsequently followed with provisional national approvals. 1917

France’s earlier objections were specific: Transport Minister Philippe Tabarot said the system could allow speeding and did not provide sufficient assurance that drivers would remain attentive, particularly in complex urban environments. 18

A two-car trial cannot establish a statistically robust collision rate on its own. Its value is narrower but important: it can independently examine operational behavior, including whether the system recognizes local road conditions, respects speed rules and maintains effective human supervision.

What an EU approval would require

A bloc-wide decision would be considered by the EU Technical Committee on Motor Vehicles. The applicable qualified-majority threshold is at least 55% of member states representing at least 65% of the EU population. 6

The five countries with provisional approval are:

• Netherlands
• Belgium
• Denmark
• Estonia
• Lithuania 1719

Those national approvals do not settle the EU-wide question. France’s position is especially consequential because it has publicly challenged whether FSD, as configured, clears the necessary safety bar. 18

Tesla’s new European safety claim

Tesla has made public an Article 39 evidence dashboard that it says supported the Dutch approval. The company says that, between April and August, FSD (Supervised) accumulated more than 100 million km across the five approved countries and had a collision rate 4.1 times lower than manually driven Tesla vehicles. 114

Tesla’s reporting methodology counts a collision as involving FSD when the feature was active at any point during the five seconds before the collision event. 13

The dashboard is useful because it exposes more of the evidence Tesla presented to regulators. But it is not an independent controlled safety study. Reuters reported that traffic-safety researchers criticized Tesla’s self-published statistical comparisons as potentially misleading, including because real-world comparisons can be affected by differences in routes, drivers, driving conditions and exposure. 2

That distinction is central to the French trial. A favorable company-reported collision ratio may be relevant evidence, but regulators still need to decide whether the evidence is independently auditable and transferable to the roads where the system would be used.

Why speed-limit compliance is a central dispute

Sweden’s transport authority has urged the EU not to approve FSD (Supervised) unless its ability to exceed detected speed limits is removed. The concern is not simply that drivers can choose an inappropriate speed; it is that an assistance system could systematically facilitate conduct that conflicts with posted legal limits. 2

France raised a related objection. Its concern was that FSD could travel at the speed of faster surrounding traffic even when that speed exceeded the posted limit. 18

The French road tests therefore offer a chance to examine a basic regulatory question: does the system consistently behave as a driver-assistance feature that follows traffic law, or does it permit speed behavior that regulators consider unacceptable for an automated driving aid?

Driver monitoring remains as important as vehicle behavior

FSD is marketed as “Supervised,” meaning the driver remains responsible for watching the road and taking over when needed. This makes the quality of driver monitoring central to any safety assessment.

France’s earlier stance explicitly questioned whether Tesla’s safeguards could guarantee driver attention in urban conditions. 18 A road test can help regulators observe the interaction between the system’s prompts, warnings and handover expectations and the real-world situations in which a driver may need to intervene.

The larger lesson is that a low reported collision rate, if validated, would not by itself answer whether a supervised system is safe enough. Regulators must also assess whether people can realistically provide the continuous oversight the system requires.

Why U.S. investigations still inform the European debate

European regulators are deciding on European road evidence, but the U.S. record remains relevant context. The National Highway Traffic Safety Administration has investigated Tesla’s FSD system over concerns including performance in reduced-visibility conditions; in March, the agency escalated a probe involving 3.2 million vehicles.

A fatal June crash in Katy, Texas initially drew attention because FSD (Supervised) had been engaged. The National Transportation Safety Board later said recovered electronic data showed that the driver manually overrode the system by fully depressing the accelerator before the vehicle struck a home.

That finding does not establish that FSD caused the crash. It does illustrate why crash attribution in supervised-driver-assistance cases can be complicated: investigators must distinguish the system’s behavior from the driver’s inputs, the road environment and the timing of any takeover.

NHTSA has also examined delays in Tesla’s mandatory crash reports involving advanced driver-assistance and autonomous-driving systems.

Transparency is part of the approval question

RDW has said it does not plan to publish key details of its FSD testing because of commercially sensitive information rules. Reuters also reported that Tesla has sought to redact basic crash details in its U.S. regulatory submissions. 19

That leaves an unavoidable policy question: how much evidence must be publicly reviewable before regulators and road users can evaluate broad deployment claims? France’s own testing may provide an additional independent data point, but it does not eliminate the need for transparent methodology and clear safety requirements.

The bottom line

France’s trial is not a fresh approval of Tesla FSD. It is an attempt to determine whether Dutch and Tesla evidence holds up under a French regulatory lens—especially on speeding, driver attention and real-road behavior.

Tesla’s 4.1-times-lower collision claim may strengthen its case, but it remains a company-generated observational comparison. Before any wider EU authorization, the core issue is whether regulators are satisfied that the system’s behavior can be independently verified, complies with road rules and keeps the human driver genuinely responsible and engaged. 1217